Regulation of the switch from early to late bacteriophage {{lambda}} DNA replication

Bacteriophage λ has played a crucial role in the development of molecular biology (Thomas, 1993 ; Gottesman, 1999 ). It is a paradigm of molecular mechanisms of many general biological processes including DNA replication and it still serves as an important model system for molecular biology research. For example, the results obtained with λ replicons were successfully employed in studies on eukaryotic DNA replication (Stillman, 1994 ), and mechanisms of replication of some animal viruses (e.g. herpes simplex virus) resemble those found in λ (Kornberg & Baker, 1992 ; Subak-Sharpe & Dargan, 1998 ).

The genome of bacteriophage λ consists of 48502 bp of double-stranded DNA (see Fig. 1). In the λ virion, this genetic material is packaged in a linear form in the head of the phage capsid. After adsorption to the surface of its host, Escherichia coli, and penetration of the phage genome into the bacterial cell, the linear DNA is immediately converted into the circular form due to single-stranded 5' extensions of 12 bases at both ends, which are complementary to each other. The ends are ligated by the host DNA ligase and, following the action of E. coli DNA gyrase, the phage genome becomes a negatively supercoiled structure (for recent reviews on λ DNA replication and phage development see Taylor & Wgrzyn, 1995 , 1998 ).

(19K): Fig. 1. (a) A genetic map of bacteriophage λ with the replication region enlarged. Positions of certain important genes are marked. The replication region of the λ genome (present also in a typical λ plasmid) is presented in the lower part of the figure. All promoters and terminators present in the replication region are indicated. Actions of proteins forming the preprimosomal complex at oriλ, and repression and stimulation of the pR promoter by Cro and DnaA, respectively, are indicated. (b) A map of λ plasmid pKB2 with marked λ genes and restriction sites used in this work during analysis of the directionality of replication by two-dimensional agarose gel electrophoresis and electron microscopy. pKB2π is a pKB2 derivative bearing a single nucleotide substitution, C to A at nucleotide 422 of the P gene (Wgrzyn et al., 1996 ).

The λ-encoded replication proteins O and P are synthesized early after phage infection, since the OP region is situated downstream of the p_R promoter and the weak t_R1 terminator (Fig. 1). The O protein is a prototype initiator protein, which binds to the origin of replication, oriλ, and directs other viral and host replication proteins (including DnaB helicase) to this site in the process of replication complex assembly. Loading of DnaB helicase (liberated from P-mediated inhibition by DnaK, DnaJ and GrpE proteins) seems to be connected with the p_R-initiated transcriptional activation of oriλ, which is stimulated by the host DnaA function (A. Wgrzyn et al., 1995a ; G. Wgrzyn et al., 1995a ; Wgrzyn et al., 1996 ; Szalewska-Paasz et al., 1998a ). This allows for the establishment of bidirectional θ replication of the λ genome, which is observed at the early stages of infection (Schnos & Inman, 1970 ). In an in vitro system reconstituted from purified proteins (and lacking HU protein), λ DNA replication was shown to be independent of transcription (Alfano & McMacken, 1989a , b ; ylicz et al., 1989 ) while proceeding unidirectionally (Mensa-Wilmot et al., 1989 ). However, addition of RNA polymerase resulted in the appearance of a significant fraction of λ DNA molecules replicating bidirectionally (Learn et al., 1993 ).

After five to six rounds of bidirectional θ replication, about 50 copies of the circular λ genome appear in the infected host cell. Then, at about 15 min post-infection, a few of these molecules start to replicate according to the σ mode (Taylor & Wgrzyn, 1995 , 1998 ). This leads to the production of long concatemers of λ DNA, up to about ten genome equivalents in length, which are cut at specific sites (called cos sites) and serve as the substrates for the phage packaging system. The mechanism of the switch from θ to σ replication has not been completely elucidated, but according to the predominant hypothesis, originally proposed by Dodson et al. (1986 ), σ may be preceded by one round of unidirectional θ replication initiated at oriλ, followed by displacement of the 5' end of the newly synthesized leading strand by its growing 3' end.

Although the mechanism of the switch from θ to σ replication has been proposed (see above), it remains unknown how this process is regulated and what triggers the change in the replication mode of λ DNA. We have found previously that the host dnaA gene function has a role in phage λ replication (G. Wgrzyn et al., 1995b ). The findings that (i) DnaA stimulates transcriptional activation of oriλ (G. Wgrzyn et al., 1995a ; Szalewska-Paasz et al., 1998a ); (ii) transcription is necessary for bidirectional θ replication of λ DNA in vitro (Learn et al., 1993 ); (iii) unidirectional θ replication is considered as a prerequisite for σ replication (Dodson et al., 1986 ); and (iv) σ replication intermediates of the phage bearing mutations in the P gene appear early after infection of the dnaA host (G. Wgrzyn et al., 1995b ; Konopa et al., 2000 ), suggested that DnaA may be a crucial factor in triggering the switch from θ to σ replication of phage λ DNA in infected E. coli cells. Therefore, the aim of this work was to investigate the mechanism of regulation of this switch in wild-type λ phage and the role of the DnaA protein in this process.

Bacterial strains, phages and plasmids.
Escherichia coli wild-type strain MG1655 (Jensen, 1993 ) and its dnaA46(ts) tnaA::Tn10 derivative (strain BM746), constructed by P1 transduction from strain BM215 (A. Wgrzyn et al., 1995b ), were used. An isogenic pair of dnaA⁺ and dnaA::cm (dnaA null) strains were also employed: this pair consists of the dnaA::cm strain TC3478 [dnaA::cm araD139 Δ(araleu) ΔlacX74 galK galU hsdR rnh-373 rpsL thi], which was described by Ingmer & Atlung (1992) , and its dnaA⁺ zid-3162::Tn10kan derivative (BM3477), which was constructed by P1 transduction from strain CAG18558, described by Singer et al. (1989 ). Phages λcIb2 and λred3cI857S7 (G. Wgrzyn et al., 1995b ) were used. λ plasmids pKB2 and pKB2π have already been described by Kur et al. (1987 ). Plasmid pTCλ1, an artificial λ plasmid bearing the p_tet promoter instead of p_R, was constructed by Herman-Antosiewicz et al. (1998a ).

One-step growth experiments.
Lytic development of bacteriophage λ was investigated by one-step growth experiments as described previously (G. Wgrzyn et al., 1995b ). Briefly, bacteria growing exponentially in LB medium at 30 °C were infected by phage λcIb2 in the presence of 3 mM NaN₃ (to prevent unsynchronized phage development) at the indicated m.o.i., and adsorption was carried out for 10 min at 43 °C. Following centrifugation, the bacterial pellet was resuspended in the same medium (with NaN₃) containing anti-λ serum and incubated for 5 min at 43 °C to neutralize unadsorbed phages. The suspension was then diluted 1000-fold with prewarmed (to 43 °C) medium (devoid of NaN₃) and aerated in a water-bath shaker. The number of infective centres was estimated by plating samples taken during the first 10 min after dilution (time 010 min). In fact, the estimated number of infective centres is the sum of the number of infected bacteria and the number of free, unadsorbed phages. However, as most of the unadsorbed phage particles were neutralized by anti-λ serum, the second value was ignored during the calculation of the burst size. The number of intracellular progeny phage was estimated by plating chloroform-treated samples of the infected culture withdrawn at different times, using strain MG1655 as a host. The burst size was calculated as the ratio of the number of progeny phages to the number of infective centres.

Density-shift experiments.
These were performed according to G. Wgrzyn et al. (1995b ). Briefly, in the first type of experiment, bacteria were grown in a light minimal medium (Wgrzyn, G. et al., 1995b ) overnight at 30 °C, and after dilution (1:50, v/v) with fresh medium the growth was continued to an OD₅₀₀ of 0·2. The bacteria were pelleted, washed with TM buffer (10 mM Tris/HCl, pH 7·2, 10 mM MgSO₄) and suspended in 0·1 vol. of this buffer. After 60 min incubation at 43 °C, [³H]thymidine-labelled phage (8·3x10^-5 c.p.m. p.f.u.^-1) was added to a m.o.i. of 10 and incubation was continued for 15 min. The suspension was sedimented, resuspended in the original volume of prewarmed (43 °C) heavy minimal medium (containing ¹⁵NH₄Cl and [¹³C]glucose instead of NH₄Cl and glucose, respectively), and further incubation was performed at 43 °C. Samples of the infected culture were withdrawn at the indicated times, and total DNA was isolated and ultracentrifuged in a CsCl density gradient as described previously (G. Wgrzyn et al., 1995b ). Fractions were collected from the bottom of the tube and the radioactivity of each fraction was measured in a scintillation counter. In the second type of experiment, the procedure was the same as described above but bacteria growing in the light medium were infected with unlabelled phage (m.o.i. of 10) and further incubation was continued in the heavy medium containing 0·1 mCi [³H]thymidine ml^-1.

Measurement of total DNA synthesis.
Synthesis of DNA in cells was investigated by measurement of incorporation of [³H]thymidine into trichloroacetic acid (TCA)-precipitable material according to the method described by Wgrzyn et al. (1991 ). Briefly, the bacterial cultures were labelled with 0·1 mCi [³H]thymidine ml^-1. Samples were withdrawn at the indicated times and transferred onto paper filters. The filters were placed immediately in 10% ice-cold TCA for 5 min, and transferred to 5% TCA for 5 min. The samples were then washed twice with 96% ethanol and dried at room temperature. The radioactivity of the samples was measured in a scintillation counter.

Two-dimensional agarose gel electrophoresis.
Analysis of replication intermediates by two-dimensional agarose gel electrophoresis was performed according to Viguera et al. (1996 ), with modifications described by rutkowska et al. (1999 ).

Electron microscopy.
Isolation of plasmid DNA was performed according to Viguera et al. (1996 ) and electron microscopy analysis of replicating plasmid DNA molecules was performed as described by Burkardt & Lurz (1984 ) and rutkowska et al. (1998 ).

Phage λ development in E. coli dnaA⁺ and dnaA hosts
Our previous studies on the role of the dnaA gene in bacteriophage λ development were performed with E. coli laboratory strains which were heavily mutagenized in their long history (G. Wgrzyn et al., 1995b ). Since such strains may also contain some as yet unidentified mutations (Glinkowska et al., 1999 ), we decided to use in most experiments of the present work E. coli wild-type strain MG1655 and its derivatives. In accordance with previous results obtained with different E. coli K-12 hosts (G. Wgrzyn et al., 1995b ; Sutton & Kaguni, 1997 ; Szalewska-Paasz et al., 1998b ) we found that the presence of the dnaA46(ts) mutation in the MG1655 genetic background allows for normal phage λ development at 43 °C (Fig. 2a). In these experiments we used phage λcIb2, which is able only to grow by the lytic pathway, to avoid a possible influence of lysogenization on the results of our studies concerning the lytic pathway (note that phage DNA replication is also a part of this developmental mode). Analogous experiments performed with phage λred3cI857S7 (used in the density-shift experiments; see below) gave very similar results, i.e. no considerable differences between the kinetics of intracellular phage development in the dnaA⁺ and dnaA46(ts) host at 43 °C (data not shown).

(14K): Fig. 2. Phage λcIb2 development in E. coli dnaA+ (MG1655; squares) and dnaA46(ts) (BM746; circles) (a), and E. coli dnaA+ (TC3478, squares) and dnaA::cm (BM3477, circles) (b) hosts at 43 °C. Phage burst size [p.f.u. per infective centre (p.f.u. IC-1)] is presented.

Since in the dnaA46(ts) host some residual activity of DnaA protein might be retained, we repeated the experiments described above except that a pair of isogenic dnaA⁺ and dnaA null (dnaA::cm) strains were used. Development of phage λcIb2 (Fig. 2b) and phage λred3cI857S7 (data not shown) was not affected by the absence of the dnaA function.

Phage λ DNA replication in E. coli dnaA⁺ and dnaA hosts
To investigate phage λ DNA replication in infected cells we employed density-shift experiments. The cells growing in a light medium were mixed with phage lysate previously labelled with [³H]thymidine, and after adsorption the cultivation was continued in a heavy medium. We found that the adsorption efficiency of the phage on both dnaA⁺ and dnaA46(ts) hosts was very similar in these conditions, with a difference below 5%. The fate of parental phage DNA was monitored by isolation of total DNA and ultracentrifugation in a caesium chloride density gradient. The recovery of labelled phage DNA was similar in both strains: 31% in dnaA⁺ and 25% in dnaA46(ts), on average. In these experiments we used the λred3 mutant to impair recombination between phage DNA molecules. When wild-type (MG1655; dnaA⁺) cells were infected, the phage DNA (originally all of it located at the full-light position, data not shown) was found in the full-light (non-replicated molecules) and heavylight (molecules after at least one replication round) positions 5 and 15 min after infection (Fig. 3a). All radioactivity moved to the heavylight fractions at 30 min, where it remained until the end of the experiment (60 min). Since we monitored the fate of parental phage DNA (labelled with [³H]thymidine), several rounds of replication proceeding by the θ mode would result in the shift of radioactivity only to the heavylight position. At later times after infection σ replication intermediates should appear and if parental DNA molecules were to enter this mode of replication, a further shift towards full-heavy position should be observed (total length of σ intermediates is up to about ten λ genomes; thus, if such a molecule contained one strand of light ³H labelled DNA of the length of one unit of the genome and the rest of the duplex DNA structure were composed of heavy nucleotides, it should be located close to the full-heavy position). However, since on average only a few of about 50 λ DNA circles that appear due to θ replication enter rolling-circle replication (see Taylor & Wgrzyn, 1995 ), no significant shift towards the full-heavy position was observed (Fig. 3a). Different results were obtained when the dnaA46(ts) bacteria were infected at 43 °C. The shift towards the full-heavy position was already observed 15 min post-infection and it was much more significant at later times (Fig. 3a). This shift strongly suggests that a larger fraction of parental phage DNA molecules enter σ replication at this temperature.

(51K): Fig. 3. The fate of DNA of infecting λ phage in E. coli dnaA+ (MG1655) and dnaA46(ts) (BM746) (a), and E. coli dnaA+ (TC3478) and dnaA::cm (BM3477) (b) hosts at 43 °C. Bacteria growing in light medium at 30 °C were infected with phage λred3cI857S7 (m.o.i. of 10) labelled with [3H]thymidine (8·3x10-5 c.p.m. p.f.u.-1), and further incubation was performed in heavy medium (containing [13C]glucose and 15NH4Cl instead of glucose and NH4Cl, respectively) at 43 °C. Samples were withdrawn at the indicated times, total DNA was isolated and centrifuged in a CsCl density gradient. Five-drop fractions were collected from the bottom of the tube and the radioactivity in each fraction was measured in a scintillation counter.

The effect of the dnaA::cm allele on phage λ DNA replication was similar to that of the dnaA46(ts) mutation at 43 °C (Fig. 3b). In these experiments, longer incubation times were employed due to considerably longer generation times of the rnh dnaA⁺ and rnh dnaA::cm strains relative to MG1655-derived strains. Similar to experiments with the dnaA46(ts) mutant, both the efficiency of phage adsorption and the recovery of parental phage DNA were comparable in dnaA⁺ and dnaA null hosts (data not shown).

Phage λ DNA synthesis in E. coli dnaA⁺ and dnaA46(ts) hosts at 43 °C
The results presented in the preceding paragraphs suggested that contrary to the λ-infected wild-type host, the early round(s) of replication of phage λ DNA in the dnaA mutants may proceed predominantly by the σ mode. Therefore, on the basis of the previously proposed model concerning the mechanism of the switch from early to late λ DNA replication (Dodson et al., 1986 ), one may speculate that early after infection of the host devoid of DnaA, λ DNA replicates according to the unidirectional θ mode, which is followed by σ replication. To test this hypothesis, we repeated the density-shift experiments except that a non-labelled phage was used. Bacteria growing in the light medium were infected with light phage and further incubation was continued in the heavy medium containing [³H]thymidine. Several rounds of θ replication of the infecting phage genome should lead to the appearance of radioactive DNA at the full-heavy position relatively shortly after infection. Up to 30 min after the shift from light to heavy medium, one replication round of the host chromosome is possible at most, thus any radioactivity found in the full-heavy fraction must come from newly synthesized λ DNA. We determined that under the conditions used in these experiments, incorporation of [³H]thymidine into TCA-precipitable material was effective during the first 15 min, and later the efficiency of incorporation dropped significantly (Fig. 4), most probably due to exhaustion of radioactive thymidine. As samples for DNA isolation and density-gradient centrifugation were withdrawn 15 and 30 min post-infection, radioactivity in the full-heavy fractions observed after the first time interval represented λ DNA molecules synthesized according to the θ mode, and that observed after the second time interval indicated the fate of λ DNA already synthesized during the first 15 min post-infection. We found considerable radioactivity in the full-heavy fractions in samples of the culture of the λ-infected dnaA⁺ host withdrawn 15 min post-infection (Table 1), indicating that several rounds of θ replication had occurred. However, low activity was observed in analogous fractions in samples of the culture of the dnaA46(ts) host infected with λ at 43 °C (Table 1). Therefore, it may be suggested that at most one round of θ replication was possible in the case of a predominant number of infecting phage DNA molecules. Effective λ DNA replication in the dnaA46(ts) mutant during the first 15 min was confirmed by monitoring the distribution of radioactivity in samples withdrawn 30 min post-infection. Significant radioactivity in the full-heavy fractions was found, perhaps representing σ replication intermediates (Table 1). Incorporation of radioactivity into DNA in the heavylight fractions (mostly E. coli chromosomal DNA) was only slightly lower in the dnaA46(ts) host than in the wild-type strain (15 min post-infection the radioactivity in heavylight fractions in the sample from the dnaA46(ts) bacteria was 71% of that found in the sample from the wild-type bacteria, and 30 min postinfection, the radioactivity in heavylight fractions in the sample from the dnaA46(ts) bacteria was 86% of that found in the sample from the wild-type bacteria). Moreover, we observed no shift of radioactivity to the full-heavy position in analogous samples from non-infected cultures (data not shown).

(12K): Fig. 4. Incorporation of [3H]thymidine (0·1 mCi ml-1) into TCA-precipitable material in E. coli dnaA+ (MG1655; squares) and dnaA46(ts) (BM746; circles) hosts growing in heavy medium at 43 °C.

Table 1. The fate of phage λ DNA synthesized during infection of E. coli dnaA+ and dnaA46(ts) hosts at 43 °C

Despite the fact that some other possible (though, in our opinion, less likely) explanations of the data presented above could be provided, these results are still compatible with the hypothesis that in the absence of DnaA function λ DNA replicates predominantly according to the unidirectional θ mode, and after one round of such replication switches to the σ mode. Thus, the next step in our studies was to test the directionality of λ DNA replication in the presence and absence of DnaA function.

Directionality of θ replication of λ DNA analysed by two-dimensional agarose gel electrophoresis
Plasmids derived from bacteriophage λ (referred to as λ plasmids) contain all the genes and regulatory sequences necessary for autonomous replication in E. coli. These plasmids replicate in the host cells by the θ mode (for reviews see Taylor & Wgrzyn, 1995 , 1998 ). Therefore, we used the λ plasmids in studies on the directionality of θ replication of λ DNA in dnaA⁺ and dnaA46(ts) strains. Wild-type λ plasmids cannot be maintained in certain dnaA(ts) mutants (including dnaA46) even at 30 °C, but derivatives bearing a mutation of the π type in the λ P gene can (Kur et al., 1987 ; Wgrzyn et al., 1996 ). Thus, in experiments with a dnaA46(ts) host we used the π mutant plasmid. Since it was demonstrated previously that even λπ plasmids can perform only one replication round in the dnaA46(ts) host after a shift from 30 °C to 43 °C (Kur et al., 1987 ), we isolated plasmid DNA for two-dimensional agarose gel electrophoresis analysis 15 min after the temperature shift.

We used E. coli dnaA⁺ and dnaA46(ts) hosts bearing plasmids pKB2 and pKB2π, respectively. Following isolation, these plasmids were digested with different restriction enzymes (shown in Fig. 1b) and analysed by two-dimensional agarose gel electrophoresis. Theoretical patterns of λ plasmid replication intermediates in the case of bidirectional and unidirectional (leftward and rightward) replication in samples digested with HindIII and BamHI, predicted by a computer method assuming that replication forks initiate synchronously and travel at the same rate (Viguera et al., 1998 ), are presented in Fig. 5(a). The results obtained and their interpretation are presented in Fig. 5(b). In the dnaA⁺ host, replication of wild-type λ plasmid (pKB2) proceeds both bidirectionally and unidirectionally. Bidirectional replication is represented by the characteristic shape of the bubble arc and the intensive descending arm of the simple-Y arc. The dots at the top and at the ascending arm of the simple-Y arc suggest rightward and leftward unidirectional replication, respectively. No significant differences in the pattern of replication intermediates were observed between DNA samples isolated from dnaA⁺/pKB2 bacteria growing at different temperatures: 30 °C, 37 °C or 43 °C (data not shown). Similar to the wild-type plasmid in the dnaA⁺ host, replication of pKB2π in dnaA46 cells at 30 °C proceeds both bidirectionally and unidirectionally. In the case of unidirectional replication, the intermediates of both leftward and rightward replication were detected. Thus, the presence of the π mutation in the plasmid and dnaA46(ts) mutation in the host genome has no significant influence on the directionality of plasmid replication at 30 °C. However, analysis of pKB2π replication intermediates isolated from the dnaA46(ts) mutant growing at 43 °C revealed that bidirectional replication was less frequent relative to that observed at 30 °C as the descending arm of the simple-Y arc is considerably less intensive.

(65K): Fig. 5. (a) Theoretical schemes of expected pictures of two-dimensional agarose gel electrophoresis of λ plasmid (pKB2 or pKB2π), digested with HindIII/BamHI, replicating according to unidirectional leftward θ mode (Uni-L), unidirectional rightward θ mode (Uni-R), bidirectional θ mode (Bi) and combination of molecules replicating according to all these modes (All). (b) Results obtained after isolation of plasmid DNA from cultures of dnaA+ (MG1655) and dnaA46 (BM746) hosts bearing pKB2 or pKB2π and growing at different temperatures.

In conclusion, replication of λ plasmids in E. coli cells proceeds both bidirectionally and unidirectionally, but in the absence of dnaA function (dnaA46 mutant at 43 °C) the bidirectional replication seems to be less frequent. This was also confirmed during the analysis of replication intermediates after digestion of plasmid DNA with other restriction enzymes: BglII and SalI, EcoRV and SalI, and NdeI (data not shown).

Electron microscopic analysis of the directionality of θ replication of λ DNA
Two-dimensional agarose gel electrophoresis demonstrated that an impaired DnaA function may result in less frequent bidirectional replication of λ plasmid DNA. To obtain quantitative data, we analysed plasmid molecules (isolated as for two-dimensional agarose gel electrophoresis) using electron microscopic techniques. Plasmid DNA molecules isolated from dnaA⁺ or dnaA46(ts) bacteria were cut with HindIII and BamHI (see Fig. 1b) and analysed by electron microscopy. Fragments of molecules containing bubbles as well as Y-shaped DNA fragments were identified and the lengths of appropriate arms were measured. For determination of the directionality of replication, the known position of oriλ was assumed to be the only possible replication start point (Fig. 6). Thus, we could determine the fraction of bidirectionally and unidirectionally replicating plasmids. We found that in the dnaA⁺ host about 40% of λ plasmid molecules replicate bidirectionally irrespective of temperature (Table 2). A similar distribution of plasmid replication intermediates was found in the dnaA46(ts) mutant growing at 30 °C, confirming the conclusion based on two-dimensional agarose gel analysis that combination of π and dnaA46(ts) mutations has little effect on directionality of λ plasmid replication at this temperature. However, among all replication intermediates found in dnaA46(ts) bacteria at 43 °C, only about 10% were derived from bidirectionally replicating plasmids (Table 2). These results are in agreement with our conclusion, based on data from two-dimensional agarose gel electrophoresis, that bidirectional DNA replication from oriλ is impaired in the absence of functional DnaA.

(176K): Fig. 6. Examples of electron microscopy analysis of λ plasmid (pKB2 or pKB2π) replication intermediates. Plasmid molecules were isolated from E. coli cells, digested with HindIII/BamHI, prepared for electron microscopy, and analysed by measuring lengths of replication bubbles and/or arms in bubble-containing and Y-shaped replication intermediates. For determination of the directionality of replication, the known position of oriλ was assumed as the only possible replication start point. Examples of bidirectionally replicating molecule (a) and unidirectionally replicating molecule (b) are presented (positions of oriλs are indicated by arrows). Bars, 250 nm.

Table 2. Electron microscopy analysis of the directionality of λ plasmid replication

DnaA affects the directionality of replication from oriλ by acting on the p_R promoter
The results described above indicated that the dnaA gene product affects the directionality of replication from oriλ directly or indirectly. Since it was demonstrated previously that DnaA directly stimulates transcription from the λ p_R promoter (Szalewska-Paasz et al., 1998a ), whose activity is necessary for transcriptional activation of oriλ, we asked whether this direct DnaA-mediated regulation is responsible for the observed effects of the dnaA46(ts) mutation on the replication from oriλ. Thus, we investigated replication of an artificial λ plasmid called pTCλ1, which bears the p_tet promoter instead of p_R (Herman-Antosiewicz et al., 1998a ). DnaA-independence of activity of the p_tet promoter has been demonstrated previously (Herman-Antosiewicz et al., 1998b ). Using both two-dimensional agarose gel electrophoresis (data not shown) and electron microscopy (Table 3) we found that inactivation of the dnaA gene product has little effect on the directionality of pTCλ1 replication (Table 3). Since the only considerable difference between pTCλ1 and standard λ plasmid replicons is in the kind of promoter located upstream of oriλ, we conclude that DnaA affects the directionality of natural λ replicons by acting on the p_R promoter.

Table 3. Electron microscopic analysis of directionality of pTCλ1 plasmid replication in E. coli dnaA46(ts) mutant

A role for the host-encoded DnaA protein in the regulation of the switch from θ to σ DNA replication of bacteriophage λ
Upon infection of its E. coli host, replication of bacteriophage λ DNA proceeds according to the θ mode and is then switched to σ mode (reviewed by Taylor & Wgrzyn, 1995 , 1998 ). Dodson et al. (1986 ) proposed that σ replication may be preceded by one round of unidirectional θ replication initiated at oriλ followed by displacement of the 5' end of the newly synthesized leading strand by its growing 3' end (Fig. 7). However, the precise mechanism regulating this switch is still not known. To find factor(s) responsible for this regulation we focused our attention on proteins that could potentially be involved in the control of directionality of DNA replication initiated at oriλ. It was previously demonstrated using an in vitro system reconstituted from purified proteins that replication of λ DNA proceeds unidirectionally in the absence of transcription, but a significant fraction of molecules replicate bidirectionally after the addition of RNA polymerase (Learn et al., 1993 ). The only transcription which is absolutely necessary for λ DNA replication in vivo and can be involved in λ DNA replication in vitro seems to be that initiated at the p_R promoter. This promoter is stimulated by the host DnaA protein (G. Wgrzyn et al., 1995a , Szalewska-Paasz et al., 1998a ). Therefore, we investigated replication of phage λ DNA in dnaA mutants.

(30K): Fig. 7. A model for the regulation of the switch from θ to σ replication of phage λ genome in wild-type bacteria (left panel) and in dnaA mutants (right panel). In a wild-type host, DnaA-mediated stimulation of transcriptional activation of oriλ allows for initiation of bidirectional θ replication of λ DNA. After five to six rounds of replication, about 50 copies of the λ genome appear. Due to presence of many DnaA binding sequences in λ DNA, cellular DnaA protein is titrated out and transcriptional activation of oriλ becomes significantly less efficient. This leads to the initiation of unidirectional θ replication, which after one round switches to the σ mode according to the mechanism previously described by Dodson et al. (1986 ), who proposed that a round of unidirectional θ replication initiated at oriλ is followed by displacement of the 5' end (dot) of the newly synthesized leading strand by its growing 3' end (arrow). In a λ-infected dnaA mutant, due to impaired stimulation of the pR promoter, unidirectional θ replication starts shortly after infection which results in the appearance of σ replication intermediates considerably earlier than in the wild-type host. For simplification of the diagram (in both panels), at the switch from θ to σ replication only the fate of the parental strand directing the synthesis of the leading strand is shown. However, one should note that there is still another copy of λ DNA after a round of replication (the copy formed due to synthesis of the lagging strand that is not shown in the diagram), which can initiate following replication round(s).

Our density-shift experiments suggested that in the absence of functional DnaA, a large fraction of parental phage λ DNA molecules enters σ replication shortly after infection. These results confirmed the previously reported electron microscopic observations that σ replication intermediates appear in λ-infected dnaA mutant cells as soon as 5 min post-infection (G. Wgrzyn et al., 1995b ; Konopa et al., 2000 ). In those studies, a λ phage bearing the mutated P gene and E. coli hosts bearing multiple mutations (apart from dnaA46) were used (G. Wgrzyn et al., 1995b ; Konopa et al., 2000 ). Here, we demonstrate that our previous results are very similar to those obtained with a host with a wild-type genetic background and λP⁺ phages. In fact, using electron microscopy we were able to observe σ replication intermediates early after infection of the strain BM764 (MG1655-derived dnaA46 mutant) by λP⁺ phage, similar to the results reported previously with more complex genetic backgrounds (G. Wgrzyn et al., 1995b ; Konopa et al., 2000 and our unpublished results).

By analysis of replication initiated at oriλ, performed using two dimensional agarose gel electrophoresis and electron microscopy, we demonstrated that while in the wild-type host and in the dnaA46(ts) mutant at 30 °C bidirectional replication is frequent, in the dnaA46(ts) host at 43 °C it is predominantly unidirectional. Since replacement of the p_R promoter with p_tet (whose activity is not dependent on DnaA) resulted in abolition of the effect of the dnaA mutation on the replication directionality, we conclude that DnaA acts in the control of directionality of replication from oriλ by regulation of transcription initiated at p_R.

Model for regulation of the switch from θ to σ DNA replication
Transcription initiated at the p_R promoter and proceeding near or through oriλ is called transcriptional activation of oriλ. This process has been known for a long time (Dove et al., 1969 ; Nijkamp et al., 1971 ) to be necessary for λ DNA replication in vivo, but its exact role was unclear until recently. The results presented in this paper, together with previously reported observations (Learn et al., 1993 ; G. Wgrzyn et al., 1995b , 1996 ), indicate that transcriptional activation of oriλ is necessary for initiation of bidirectional replication from this region. This led us to propose a mechanism of regulation of the switch from θ to σ replication of bacteriophage λ DNA. Since DnaA positively regulates transcription from the p_R promoter (Szalewska-Paasz et al., 1998a ), its activity is necessary for frequent initiation of bidirectional replication from oriλ in E. coli. Early after infection, in spite of the presence of many DnaA boxes in the E. coli chromosome, there is a sufficient concentration of free DnaA molecules in the cell to stimulate bidirectional θ replication, as maximal activation of p_R by DnaA occurs at relatively low concentrations of this protein (Szalewska-Paasz et al., 1998a ). However, after a few rounds of bidirectional θ replication, many copies of the λ genome appear (about 50 copies after five to six replication rounds). Since it was previously demonstrated by Szalewska-Paasz et al. (1998c ) that there are many DnaA binding sites in λ DNA, the DnaA protein may be titrated out. This should lead to an inefficient transcriptional activation of oriλ resulting in unidirectional θ replication followed by the σ replication mode. In dnaA mutants, weak transcription from the p_R promoter allows predominantly unidirectional θ replication and a switch to σ replication shortly after infection. This model is presented schematically in Fig. 7. If this model is correct, the DnaA box sequences responsible for titrating out DnaA molecules after several rounds of bidirectional θ replication of the λ genome in the wild-type host should be of weak affinity to this protein. This would allow efficient transcriptional activation of oriλ and bidirectional θ replication at the beginning of the infection cycle. In accordance with this prediction, it was reported that most DnaA boxes present in the λ genome are weak ones (Szalewska-Paasz et al., 1998c ).

Better & Freifelder (1983 ) reported that σ replication intermediates can be also occasionally found at early times after infection of wild-type hosts by λ phage. They found that θ replication is predominant, but not exclusive, at early stages of infection and σ replication is predominant, but not exclusive, at late stages of infection. This is also compatible with the results presented in this report and with our model of the switch from θ to σ replication, as we have demonstrated that in the presence of DnaA function the replication of λ plasmids proceeds bidirectionally and unidirectionally, and in the absence of DnaA function the replication is predominantly, but not exclusively, unidirectional. In fact, after inactivation of DnaA we observed (using electron microscopy) a decrease in the fraction of bidirectionally replicating λ DNA from 40% to 10%, whereas the dnaA46(ts) mutation at 43 °C was responsible for a huge increase in σ-type replication intermediates as revealed by density-shift experiments. However, one should note that the electron microscopy studies were performed with plasmids occurring in many copies per cell, whereas in the density shift experiments only a few phages infected one cell. Moreover, whole population of λ plasmid molecules was investigated by electron microscopy, whereas the fate of only parental phage λ DNA molecules was monitored in density-shift experiments.

A physiological role for the switch from θ to σ DNA replication
It is also interesting to consider a physiological role for a DnaA-regulated switch from θ to σ replication of phage λ DNA. Since efficient growth of λ phage was observed in the dnaA46(ts) mutant at 43 °C, i.e. under conditions in which σ replication starts early after infection, it raises the question as why should θ replication occur at early stages of infection in a wild-type host? The physiological role of θ replication should be the production of many copies of λ DNA to be used as templates for high level of expression of phage genes employed during lytic development. If this is true, the amount of λ DNA produced by σ replication is high enough for expression of phage proteins in amounts sufficient to support normal lytic development. However, it is worth noting that studies on λ development, which demonstrated efficient production of phage progeny in dnaA mutants (G. Wgrzyn et al., 1995b ; Sutton & Kaguni, 1997 ; Szalewska-Paasz et al., 1998b ; Glinkowska et al., 1999 ; this work), were carried out under standard laboratory conditions (i.e. in LB medium with good aeration, etc.) that support high E. coli growth rates. It was demonstrated recently that under these conditions, in a wild-type host, phage λ produces at least some proteins in excess of those needed for its effective propagation: thus efficient growth of the phage may be achieved even under less favourable conditions (Gabig et al., 1998 ). This phage developmental strategy may be responsible for efficient λ growth in dnaA mutants, in which expression of phage genes is expected to be decreased relative to the wild-type host. If this is true, one might expect an inhibition of λ lytic development in a dnaA mutant growing in poor media. Indeed, it has been reported that phage λ lytic growth is less efficient in wild-type cells cultivated in minimal media, and it may be completely inhibited in slowly growing dnaA46(ts) mutants at 43 °C (Wgrzyn et al., 2000 ).

We acknowledge the invaluable contribution of Karol Taylor at the beginning of this project. We thank Joanna Potrykus and Katarzyna Potrykus for critical reading of the manuscript. This work was supported by the USAPoland Maria SklodowskaCurie Joint Fund II (grant MEN/HHS-96-255), the Polish State Committee for Scientific Research (project 6 P04A 016 16), the Volkswagen Foundation (grant I/74 639), the University of Gdask (grant BW/1190-5-0145-9), the Spanish Dirección General de Enseñanza Superior (grant PM95/0016), the Spanish Fondo de Investigación Sanitaria (grant 96/0470) and the Comunidad de Madrid (grant 08.6/0016/1997). We acknowledge financing of a short term fellowship of S.B. in the CSIC (Madrid) by EMBO (award no. ASTF9021). G.W. acknowledges also financial support from the Foundation for Polish Science (subsydium 14/2000).